The ALK Inhibitor Ceritinib Overcomes Crizotinib Resistance in Non–Small Cell Lung Cancer

Published OnlineFirst March 27, 2014; DOI: 10.1158/2159-8290.CD-13-0846

RESEARCH ARTICLE

Luc Friboulet 1 , 2, Nanxin Li 3 , Ryohei Katayama 1 , 2 , 4, Christian C. Lee 3 , Justin F. Gainor 1 , 2, Adam S. Crystal 1 , 2, Pierre-Yves Michellys 3 , Mark M. Awad 1 , 2, Noriko Yanagitani 5 , Sungjoon Kim 3 , AnneMarie C. Pferdekamper 3 , Jie Li 3, Shailaja Kasibhatla 3 , Frank Sun 3 , Xiuying Sun 3, Su Hua 3, Peter McNamara 3 , Sidra Mahmood 1 , 2, Elizabeth L. Lockerman 1 , 2, Naoya Fujita 4 , Makoto Nishio 5 , Jennifer L. Harris 3 , Alice T. Shaw 1 , 2, and Jeffrey A. Engelman 1 , 2

ABSTRACT Non–small cell lung cancers (NSCLC) harboring anaplastic lymphoma kinase (ALK ) gene rearrangements invariably develop resistance to the ALK tyrosine kinase inhibitor (TKI) crizotinib. Herein, we report the fi rst preclinical evaluation of the next-generation ALK TKI, ceritinib (LDK378), in the setting of crizotinib resistance. An interrogation of in vitro and in vivo models of acquired resistance to crizotinib, including cell lines established from biopsies of patients with crizotinib-resistant NSCLC, revealed that ceritinib potently overcomes crizotinib-resistant mutations. In particular, ceritinib effectively inhibits ALK harboring L1196M, G1269A, I1171T, and S1206Y mutations, and a cocrystal structure of ceritinib bound to ALK provides structural bases for this increased potency. However, we observed that ceritinib did not overcome two crizotinib-resistant ALK mutations, G1202R and F1174C, and one of these mutations was identifi ed in 5 of 11 biopsies from patients with acquired resistance to ceritinib. Altogether, our results demonstrate that ceritinib can overcome crizotinib resistance, consistent with clinical data showing marked effi cacy of ceritinib in patients with crizotinib-resistant disease.

SIGNIFICANCE: The second-generation ALK inhibitor ceritinib can overcome several crizotinib- resistant mutations and is potent against several in vitro and in vivo laboratory models of acquired resistance to crizotinib. These ﬁ ndings provide the molecular basis for the marked clinical activity of ceritinib in patients with ALK -positive NSCLC with crizotinib-resistant disease. Cancer Discov; 4(6); 662–73. ©2014 AACR.

See related commentary by Ramalingam and Khuri, p. 634.

Authors’ Afﬁ liations: 1Massachusetts General Hospital Cancer Center; Corresponding Authors: Jeffrey A. Engelman, Massachusetts General 2 Department of Medicine, Harvard Medical School, Boston, Massachu- Hospital Cancer Center, CNY 149, 13th Street, Charlestown, MA 02129. setts; 3 Genomics Institute of the Novartis Research Foundation, San Phone: 617-724-7298; Fax: 617-724-9648; E-mail: jengelman@partners Diego, California; 4 Cancer Chemotherapy Center and 5Cancer Institute .org; Alice T. Shaw, [email protected] ; and Jennifer L. Harris, jharris@ Hospital, Japanese Foundation for Cancer Research, Tokyo, Japan gnf.org. Note: Supplementary data for this article are available at Cancer Discovery doi: 10.1158/2159-8290.CD-13-0846 Online (http://cancerdiscovery.aacrjournals.org/). ©2014 American Association for Cancer Research. L. Friboulet, N. Li, and R. Katayama contributed equally to this work.

662 | CANCER DISCOVERYJUNE 2014 www.aacrjournals.org

INTRODUCTION Table 1. Ceritinib is a potent ALK inhibitor Chromosomal rearrangements of anaplastic lymphoma kinase (ALK ) are detected in 3% to 7% of non–small cell lung cancers (NSCLC; refs. 1, 2). These rearrangements result in GI50 (nmol/L) constitutively active ALK fusion proteins with potent trans- Crizotinib Ceritinib Fold forming activity ( 2, 3 ). Lung cancers with ALK rearrangements are highly sensitive to ALK tyrosine kinase inhibition, ALK enzymatic assay 3 0.15 20 underscoring the notion that such cancers are addicted to H2228 107 3.8 28 ALK kinase activity. On the basis of early-phase studies, the H3122 245 6.3 39 multitargeted tyrosine kinase inhibitor (TKI) crizotinib was approved by the FDA in 2011 to treat patients with advanced NOTE: GI 50 values for in vitro ALK enzymatic assay or H3122 and H2228 cell survival assay for crizotinib and ceritinib are shown. NSCLC harboring ALK rearrangements (1 ). However, despite a high response rate of 60% in ALK -rearranged NSCLC, most patients develop resistance to crizotinib, typically within 1 to 2 years. RESULTS Studies of ALK -rearranged lung cancers with acquired resistance to crizotinib have identifi ed ALK fusion gene Ceritinib Exhibits Potent Activity in amplifi cation and secondary ALK tyrosine kinase (TK) Crizotinib-Naïve ALK -Positive NSCLC Models domain mutations in about one third of cases (4–6 ). To In vitro enzymatic studies revealed that ceritinib was date, seven different acquired resistance mutations have been approximately 20-fold more potent against ALK than cri- identifi ed among crizotinib-resistant patients. The most fre- zotinib ( Table 1 ). Similarly, ceritinib was more potent than quently identifi ed secondary mutations are L1196M and crizotinib against two ALK -rearranged lung cancer cell lines, G1269A. In addition to these mutations, the 1151T-ins, H3122 and H2228 ( Fig. 1A and B , Table 1 ). Accordingly, L1152R, C1156Y, G1202R, and S1206Y mutations have also ceritinib led to suppression of ALK phosphorylation as well been detected in crizotinib-resistant cancers ( 4 , 6–10 ). In as the downstream PI3K–AKT, MEK–ERK, and mTOR signal- approximately one third of crizotinib-resistant tumors, there ing pathways at lower doses than crizotinib ( Fig. 1C and D ). is evidence of activation of bypass signaling tracts such as To further assess the cellular specifi city of ceritinib, we deter-

EGFR or c-KIT ( 6 , 9 ). In the remaining one third of crizo- mined the GI50 (concentration needed to reduce the growth of tinib-resistant tumors, the resistance mechanisms remain to treated cells to half that of untreated cells) of ceritinib against be identifi ed. a panel of tumor cell lines bearing different oncogenic driv- Next-generation ALK inhibitors with improved potency ers. Whereas ceritinib was potent against the two lung cancer and selectivity compared with crizotinib have been devel- cell lines with ALK rearrangements, it was not potent against oped to overcome crizotinib resistance in the clinic. We pre- NSCLC or breast cancer cell lines driven by KRAS, EGFR, PI3K, > μ viously evaluated the ability of several ALK TKIs (TAE684, or HER2, with GI50 s 1 mol/L (Supplementary Fig. S1A). AP26113, ASP3026, and CH5424802) to inhibit ALK activ- We next compared the effi cacy of ceritinib and crizotinib in ity in models harboring different ALK secondary muta- vivo using treatment-naïve H2228 xenograft models ( Fig. 1E ). tions (6 , 11 ). These studies revealed variable sensitivity to Tumor-bearing animals were treated with either high-dose cri- these ALK inhibitors depending on the specifi c resistance zotinib (100 mg/kg) or ceritinib (25 mg/kg or 50 mg/kg) once mutation present. For example, the gatekeeper L1196M daily for 14 days. Both crizotinib (100 mg/kg) and ceritinib mutation was sensitive to TAE684, AP26113, and ASP3026, (25 and 50 mg/kg) were well tolerated in this study (Supple- whereas 1151T-ins conferred resistance to all next-gener- mentary Fig. S1B). As expected, marked tumor regression was ation ALK TKIs. Ceritinib is an ATP-competitive, potent, observed in all groups during the treatment. After treatment and selective next-generation ALK inhibitor (12 ). The kinase was stopped, the animals were monitored for tumor progres- selectivity has been tested in a cellular proliferation assay sion. Although recurrent tumors were detected within 11 days against 16 different kinases, and aside from ALK, no inhi- of drug withdrawal in mice treated with crizotinib, mice treated bition below 100 nmol/L was observed (12 ). In the phase I with ceritinib at 50 mg/kg remained in complete remission with study of ceritinib in ALK-positive NSCLC, marked antitu- no discernible tumor growth for 4 months. In the mice treated mor activity has been observed in both crizotinib-relapsed with ceritinib at 25 mg/kg, tumor regrowth was observed in 4 and crizotinib-naïve patients (13, 14). On the basis of this of 8 animals after 1 month, whereas complete remission was impressive clinical activity, ceritinib received FDA approval maintained in the other 4 animals for 4 months. Thus, ceritinib on April 29, 2014. had more durable antitumor activity than crizotinib, even after Herein, we present the fi rst report examining the activity the drugs were discontinued. It is also worth noting that the of ceritinib in preclinical models of ALK -positive lung can- exposure of crizotinib at 100 mg/kg is approximately 3-fold to cer with acquired resistance to crizotinib, as well as an early 5-fold greater than the exposures achieved at the human maxi- biologic insight into mechanisms of resistance to ceritinib mum tolerated dose (MTD; 250 mg, twice a day; ref. 15 ) and arising in patients. that ceritinib at 25 to 50 mg/kg is predicted to be achievable at

JUNE 2014CANCER DISCOVERY | 663

RESEARCH ARTICLE Friboulet et al.

A H3122 B H2228

100 100 Crizotinib Crizotinib Ceritinib Ceritinib

50 50 % Cell viability% Cell % Cell viability% Cell

0 0 0 1 10010 1,000 10,000 0 1 10 100 1,000 1,0000 Drug concentration (nmol/L) Drug concentration (nmol/L)

C H3122 D H2228 Crizotinib Ceritinib Crizotinib Ceritinib

nmol/L: 0 30 100 300 1,000 0 30 100 300 1,000 nmol/L: 0 30 100 300 1,000 0 30 100 300 1,000

pALK pALK ALK ALK pAKT pAKT AKT AKT pERK pERK ERK ERK pS6 pS6 S6 S6 GAPDH GAPDH

E H2228 tumor models 1,200 Vehicle, n = 8 Crizotinib 100 mg/kg, n = 8

) 1,000 3 Ceritinib 25 mg/kg, n = 8 800 Ceritinib 50 mg/kg, n = 8

600

400

Tumor volume(mm 200

0 0 15 30 45 60 75 90 105 120 135 150 Days after treatment

Figure 1. Ceritinib is a potent ALK inhibitor in crizotinib-naïve models. A and B, cell survival assay of H3122 (A) and H2228 (B) cells treated with the indicated doses of crizotinib or ceritinib for 72 hours. Cell survival was assayed by CellTiter-Glo. C and D, H3122 (C) and H2228 (D) cells were treated with the indicated concentrations of crizotinib or ceritinib for 6 hours. Lysates were probed with antibodies directed against the speciﬁ ed proteins. E, SCID beige bearing H2228 cells were administered crizotinib or ceritinib orally once daily for 14 days. The arrow indicates when treatments were stopped, and tumor growth was monitored in animals up to 4 months. Tumor volumes, mean ± SD (n = 8). p, phosphorylated.

the human MTD (750 mg every day). We also evaluated the efﬁ - Ceritinib Is Active against Patient-Derived Cell cacy of ceritinib in a primary explant model derived from a cri- Lines from Crizotinib-Resistant Cancers with and zotinib-naïve NSCLC tumor MGH006 (6 ). Treatment of these without Resistant Mutations mice with 25 mg/kg ceritinib also led to tumor regressions (Supplementary Fig. S1C). Altogether, these data demonstrate To investigate the activity of ceritinib against crizotinib- that ceritinib is potent against crizotinib-naïve ALK -rearranged resistant mutations, we used crizotinib-resistant cell line cell lines and tumor models in vivo and in vitro. models harboring the two most common EML4–ALK

664 | CANCER DISCOVERYJUNE 2014 www.aacrjournals.org

Ceritinib and Crizotinib Resistance RESEARCH ARTICLE

ABH3122 CR1 (L1196M) Crizotinib Ceritinib 0 30 100 300 nmol/L: 0 30 100 300 1,000 1,000 GI (nmol/L) 50 pALK ALK ALK Cell line mutation Crizotinib Ceritinib Fold pAKT H3122 CR1 L1196M 2884 230 13 AKT MGH021-4 G1269A 500 80 6 pERK MGH045 L1196M 891 25 36 ERK GAPDH C MGH021-4 (G1269A) D MGH045 (L1196M) Crizotinib Ceritinib Crizotinib Ceritinib 100 0 300 1,000 1,000 0 30 nmol/L: 30 100 300 0 300 nmol/L: 30 100 1,000 0 30 100 300 1,000 pALK pALK ALK ALK pAKT pAKT AKT AKT pERK pERK

ERK ERK

GAPDH GAPDH

E MGH045 (L1196M)

2,500 Vehicle, n = 6 Crizotinib 100 mg/kg, n = 6 n 2,000 Ceritinib 25 mg/kg, = 6 ) 3

1,500

1,000 Figure 2. EML4–ALK L1196M and G1269A mutations are sensitive to ceritinib in vitro. A, GI 50 values of cell survival assay for crizotinib or ceritinib in cell lines harboring L1196M or G1269A crizotinib Tumor volume (mm 500 resistant mutations. B to D, H3122 CR1 (B), MGH021-4 (C), and MGH045 (D) cells were treated with the indicated concentrations of crizotinib or ceritinib for 6 hours. Lysates were probed with antibod- 0 ies directed against the indicated proteins. E, nude mice bearing the 0 7 14 21 28 MGH045 cell line were treated with 100 mg/kg crizotinib or ceritinib Days of treatment 25 mg/kg. Tumor volumes, mean ± SD ( n = 6). p, phosphorylated.

mutations, L1196M and G1269A. We have previously tion ( 5 ). This clone, therefore, represents an early generation

described the H3122 CR1 crizotinib-resistant cell line, which of the patient-derived cell line. The GI50 values of ceritinib developed resistance in vitro by chronic exposure to crizotinib. against all of these resistant cell lines were decreased 6-fold This cell line harbors both the L1196M EML4–ALK gate- to 36-fold compared with crizotinib ( Fig. 2A and Supplemen- keeper mutation and ampliﬁ cation of the EML4–ALK allele tary Fig. S2A–S2C). Accordingly, phosphorylation of ALK (11 ). In addition, we also examined two novel cell lines estab- and downstream ERK and AKT were more effectively sup- lished from biopsies of patients whose ALK -rearranged lung pressed by lower doses of ceritinib compared with crizotinib cancers had become resistant to crizotinib in the clinic. These (Fig. 2B–D). two patient-derived resistant lines, MGH045 and MGH021- To further assess the activity of ceritinib against crizotinib- 4, harbor the L1196M and G1269A mutations, respectively. resistant ALK -positive tumors in vivo , we examined the efﬁ - The MGH021-4 line is a clonal cell line established from cacy of ceritinib against xenografts derived from MGH045 MGH021, a tumor harboring both 1151T-ins and G1269A cells that harbor the L1196M resistance mutation. As shown mutations; MGH021-4 cells harbor only the G1269A muta- in Fig. 2E, treatment of MGH045 tumor-bearing mice with

JUNE 2014CANCER DISCOVERY | 665

RESEARCH ARTICLE Friboulet et al.

A Pre-crizotinib Post-crizotinib

MGH051 (WT) B C Crizotinib Ceritinib MGH051 (WT) 0 30 300 1,000 30 100 300 1,000 nmol/L: 100 0 100 pALK Crizotinib GI50 = 62 nmol/L Ceritinib GI = 2.6 nmol/L 50 ALK

pAKT

50 AKT pERK % viability Cell ERK

0 pS6 0 1 10010 1,000 10,000 S6 Drug concentration (nmol/L) GAPDH

Figure 3. Ceritinib is active in ALK wild-type (WT) crizotinib-resistant cell line. A, abdominal computed tomography (CT) images of patient MGH051 before treatment with crizotinib and after 11 weeks of crizotinib. Several new hepatic metastases (yellow arrows) were detectable after crizotinib treatment consistent with disease progression. A repeat biopsy of a hepatic metastasis was performed within 2 weeks of crizotinib discontinuation. B, MGH051 cells were treated with the indicated doses of crizotinib or ceritinib for 7 days. After the incubation, the cell survival was assayed by CellTiter-Glo. C, MGH051 cells were treated with the indicated concentrations of crizotinib or ceritinib for 24 hours. Lysates were probed with antibodies directed against the indicated proteins.

low-dose ceritinib (25 mg/kg) was more effective than with crizotinib. Assessment of the biopsy sample revealed no ALK high-dose crizotinib in controlling tumor growth. These data mutations or gene amplifi cation. The cell line derived from demonstrate that ceritinib is active against cancers derived the biopsy also did not harbor any ALK resistance mutations. from patients with acquired resistance to crizotinib and is This resistant cell line was highly sensitive to ceritinib in vitro, more potent than crizotinib against ALK -rearranged cancers and, surprisingly, the MGH051 cell line was also sensitive to harboring the L1196M and G1269A resistance mutations. crizotinib ( Fig. 3B ). Accordingly, phosphorylation of ALK The ongoing clinical trial of ceritinib demonstrates that and downstream AKT and ERK was effi ciently suppressed crizotinib-resistant ALK -positive tumors, including tumors by crizotinib and ceritinib (Fig. 3C). These data suggest that without ALK mutation or gene amplifi cation, are respon- cancers with acquired resistance to crizotinib without ALK - sive to ceritinib treatment (13 ). This raises the possibility resistant mutations may remain sensitive to ALK inhibition that many of these resistant tumors may develop because (please see “Discussion”). of inadequate target suppression. We investigated the effi - cacy of crizotinib and ceritinib against a crizotinib-resistant Assessment of Ceritinib Activity against ALK-positive cell line without ALK resistance mutations, a Panel of ALK Mutations MGH051. As shown in Fig. 3A , this cell line was derived To systematically assess the potency of ceritinib against from a biopsy of a liver lesion that developed in a patient on ALK resistance mutations, we used Ba/F3 cells engineered

666 | CANCER DISCOVERYJUNE 2014 www.aacrjournals.org

Ceritinib and Crizotinib Resistance RESEARCH ARTICLE

to express wild-type EML4–ALK or one of the nine dif- in cocrystal structures. Interestingly, positions T1151 and ferent resistance mutations. In this system, ceritinib was F1174 in ALK have been previously identifi ed as sites of approximately 10-fold more potent against wild-type EML4– activating gain-of-function mutations in neuroblastoma ALK than crizotinib. Whereas all these secondary muta- (17 ). Although diffi cult to predict without structural and tions induced crizotinib resistance, ceritinib was potent in biochemical analyses of these mutants, T1151 is adjacent inhibiting the growth of Ba/F3 cells expressing four of the to the catalytically important K1150, and insertion at this resistance mutations, including L1196M, G1269A, S1206Y, position, along with the F1174C, L1152P, and C1156Y and I1171T ( Fig. 4A ; Supplementary Fig. S3; Supplementary inhibitor-resistant mutants, likely infl uences αC-helix Table S1). However, C1156Y, G1202R, 1151T-ins, L1152R, mobility and conformational dynamics of the catalytic and F1174C mutations also conferred resistance to ceritinib, domain. Previously reported structures of nonphospho- although ceritinib was still more potent than crizotinib rylated ALK in the apo, ADP, and inhibitor-bound forms against these mutations. Thus, the most common crizo- suggest the ALK catalytic domain structure possesses a tinib-resistant mutations were substantially more sensitive “DFG-in” conformation (18 ) with a unique activation loop to ceritinib than crizotinib, whereas less common resist- conformation. It is conceivable that these mutations desta- ance mutations conferred resistance to both crizotinib and bilize the ALK conformation and shift the conformational ceritinib. equilibrium toward those that are no longer able to bind the inhibitor. It is also possible that these mutations could Structural Basis for Increased Potency of Ceritinib decrease the K m for ATP, rendering ceritinib/crizotinib a less against ALK Crizotinib-Resistant Mutations effective ATP competitive inhibitor. To glean insights into the structural basis for the ability of ceritinib to maintain activity toward select crizotinib- Crizotinib-Resistant Tumors Harboring EML4–ALK resistant mutants, the structure of the ALK catalytic domain Wild-Type, I1171T, or C1156Y Mutations Are complexed with ceritinib was determined ( Fig. 4B ; PDB Sensitive to Ceritinib In Vivo 4MKC) and compared with the structure of the ALK cata- To evaluate the activity of ceritinib against crizotinib- lytic domain bound to crizotinib (Fig. 4C; PDB 2XP2; ref. resistant tumors in vivo, crizotinib-resistant H2228 16 ). As shown in Fig. 4A , ceritinib retains potency toward xenograft tumors were generated by treatment with esca- the most common G1269A and L1196M crizotinib-resist- lating doses of crizotinib (from 50 to 100 mg/kg). Tumors ant mutants. The cocrystal structure reveals that G1269 is that progressed during treatment with 100 mg/kg crizotinib situated just proximal to D1270 of the activation loop DFG- were analyzed for resistance mechanisms. Typical tumor motif. Although mutation to Ala in the G1269A mutant responses and resistance are shown for 3 animals in Sup- would not be predicted to present any steric obstruction to plementary Fig. S4, and are representative of the 80 animals ceritinib binding, it would be predicted to introduce a steric used in this study. To determine mechanisms of resistance clash to crizotinib binding due to the proximity of the phe- to crizotinib, we sequenced the ALK kinase domain of all 80 nyl ring of crizotinib. The Cl moiety of the pyrimidine hinge- tumors and identifi ed three distinct resistance mutations binding core of ceritinib is juxtaposed with the L1196 side in six tumors. The G1202R, C1156Y, and I1171T muta- chain and participates in a hydrophobic interaction with tions were detected in three, two, and one resistant tumors, the Leu side chain. In the L1196M mutant, the Cl moiety respectively. Of these three mutations, G1202R and C1156Y of ceritinib can interact with Met, which may compensate have been previously reported in patients with NSCLC who for the loss of interaction between Cl and the Leu side chain relapsed on crizotinib (6, 7). Interestingly, I1171T has not in wild-type ALK. In contrast, introduction of a Met at the yet been reported from crizotinib-resistant patients but was gatekeeper position 1196 likely adversely affects crizotinib identifi ed in an in vitro mutagenesis screen for resistance binding through both steric interference and unfavorable mutations (19 ). interactions with the 2-amino substituent of the pyridinyl The efficacy of ceritinib was tested against these crizo- hinge-binding core and the methyl substituent of the alkoxy tinib-resistant H2228 xenograft tumor models as well as moiety of crizotinib. These structural fi ndings are in agree- one of the resistance models that did not harbor a resist- ment with the increased potency of ceritinib versus crizo- ance mutation nor ALK amplification (data not shown). tinib against these resistance mutations. Although each was resistant to crizotinib at 100 mg/kg, In contrast with G1269A and L1196M mutations, ceritinib ceritinib suppressed tumor growth in multiple resistance is not potent against the G1202R crizotinib-resistant muta- models ( Fig. 5A–D ). In the wild-type and I1171T resist- tion (Fig. 4A). The crystal structure reveals that mutation ant models, ceritinib demonstrated impressive antitumor of G1202 to a larger, bulky, and charged side chain would activity, whereas it was less active in the C1156Y-resist- be incompatible with ceritinib or crizotinib ALK binding ant model and was inactive against the G1202R-resistant due to steric hindrance (6 ). This steric obstruction leads model. These data are consistent with the Ba/F3 models

to a loss in potency as reﬂ ected by the shift in IC50 values in which ceritinib was more potent against I1171T than observed for ceritinib and crizotinib. In contrast with the the C1156Y and G1202R mutants (Fig. 4A). The studies G1202R mutation, the T1151 insertion, L1152P, C1156Y, shown herein provide evidence that ceritinib can overcome and F1174C inhibitor–resistant mutants all map to the resistance in vivo, especially in tumors harboring wild-type, N-terminal lobe of the ALK catalytic domain and ﬂ ank L1196M, or I1171T ALK fusions at a dose that is predicted opposing ends of the αC-helix. The locations of these to be achievable in humans. Of note, it is rather interest- mutants do not directly contribute to inhibitor binding ing that ceritinib overcame crizotinib resistance in the

JUNE 2014CANCER DISCOVERY | 667

RESEARCH ARTICLE Friboulet et al.

A Crizotinib Ceritinib 1,000

100

values (nmol/L) values 10 50 IC

Parental WT V1 WT V3 I1171T V1 1171T V3 S1206Y V1 F1174C V3I L1152P V3 0 1151Tins V1G1202R V1 G1269A V1L1196M V1 G1269S V3L1196M V3C1156Y V3 B

CI N C1156 T1151 L1152 O O HN N NH O S L1196 L1196

F1174 G1269 G1202 G1269 S1206 N H

Ceritinib

F C1156 CI T1151 NH2 L1152 O L1196 L1196 N CI F1174 G1269 G1269

G1202 N N S1206

Crizotinib

Figure 4. Ba/F3 models of ALK-crizotinib–resistant mutations. A, IC50 of ceritinib across different Ba/F3 cell lines expressing wild-type or mutated ALK TK and including parental, IL3-dependent Ba/F3 cells are shown. B and C, ALK-resistant mutations mapped onto ALK/ceritinib (PDB 4MKC; B) and ALK/crizotinib (PDB 2XP2; C) cocrystal structures. β-Strand secondary structural elements of the N-terminal lobe and the αC-helix of the N-terminal lobe are shown in orange and purple, respectively. Helical structural elements of the C-terminal lobe are shown in blue. Residues of the activation loop (A-loop) and catalytic loop are shown in red and orange, respectively. Residues involved in resistant mutations are depicted as green spheres. Inhibitor molecules are depicted as stick representations with carbons colored yellow and cyan for crizotinib and ceritinib, respectively. Nitrogen is colored dark blue, oxygen is colored red, and chlorine green for both inhibitors. Fluoride is colored white (crizotinib) and sulfur atoms are colored yellow (ceritinib). Transparent surfaces for the inhibitors are displayed. Zoomed-in view boxes for G1269 and L1196 residues are shown. Figures were rendered with MacPymol (The PyMOL Molecular Graphics System, Version 1.4 Schrödinger, LLC).

668 | CANCER DISCOVERYJUNE 2014 www.aacrjournals.org

Ceritinib and Crizotinib Resistance RESEARCH ARTICLE

A H2228 crizotinib-resistant tumor B H2228 crizotinib-resistant tumor model with EML4–ALK WT model with EML4–ALK I1171T mutation

1,000 1,500

800 ) 3 ) 3 1,000 600

400 500 Tumor volume (mm Tumor volume (mm 200

0 0 0 5 10 15 20 25 0 5 10 15 Days of treatment Days of treatment Vehicle, n = 8 Vehicle, n = 6 Crizotinib 100 mg/kg, n = 8 Crizotinib 100 mg/kg, n = 6 Ceritinib 50 mg/kg, n = 8 Ceritinib 50 mg/kg, n = 6

C H2228 crizotinib-resistant tumor D H2228 crizotinib-resistant tumor model with EML4–ALK C1156Y mutation model with EML4–ALK G1202R mutation 1,000 1,500

800 ) ) 3 3 1,000 600

400 500 Tumor volume (mm Tumor volume (mm 200

0 0 0 5 10 15 0 5 10 15 20 Days of treatment Days of treatment Vehicle, n = 8 Vehicle, n = 5 Crizotinib 100 mg/kg, n = 8 Crizotinib 100 mg/kg, n = 5 Ceritinib 50 mg/kg, n = 8 Ceritinib 50 mg/kg, n = 5

Figure 5. EML4–ALK C1156Y, I1171T, G1202R mutations’ sensitivity to ceritinib. A to D, SCID beige mice bearing H2228 crizotinib-resistant tumors EML4–ALK wild-type (WT; A), I1171T (B), C1156Y (C), or G1202R (D) were treated with 100 mg/kg crizotinib or 50 mg/kg ceritinib once daily for 12 to 22 days. Tumor volumes, mean ± SD ( n = 5–8).

tumor that did not harbor an ALK resistance mutation, responses, tumors do develop resistance. We have now biop- as this recapitulates observations in the clinic and with sied 11 cancers with acquired resistance to ceritinib (two of the patient-derived cell line shown in Fig. 3 (please see which were from different sites from the same patient). As “Discussion”). shown in Fig. 6A , fi ve of these biopsies revealed the development of mutations at either G1202 or F1174 in the ceritinib- Acquired Resistance to Ceritinib in Patients resistant cancers. In the patient JFCR021, who had two sites Ceritinib has demonstrated impressive activity in the clinic of disease biopsied, two different ceritinib-resistant mutations in crizotinib-resistant patients ( 13 ). However, similar to were identifi ed, underscoring the heterogeneity of resistance other targeted therapy successes, despite initial and durable mechanisms that can be identifi ed in a single patient (6 ). Of

JUNE 2014CANCER DISCOVERY | 669

RESEARCH ARTICLE Friboulet et al.

A EML4–ALK sequence EML4–ALK sequence Patient ID at crizotinib at ceritinib resistance resistance

MGH011 S1206Y G1202R

MGH015 WT WT

MGH023 WT F1174C

MGH034 WT WT

MGH049 WT WT

MGH051 WT G1202R

MGH057 N/A WT

MGH061 WT WT

JFCR013 N/A WT

F1174V (left lung) and JFCR021 G1269A (right lung) G1202R (right lung)

B MGH011 lung CT scan

Baseline After 8 weeks After 34 months After 12 weeks After 15 months of crizotinib of crizotinib of ceritinib of ceritinib EML4–ALK sequence: WT S1206Y G1202R

Figure 6. Ceritinib-resistant tumors acquired mutations at positions G1202 or F1174. A, ALK mutational status in ceritinib-resistant patient tumors before and after ceritinib treatment. B, thoracic computed tomography (CT) images of patient MGH011 during crizotinib or ceritinib treatments. Sites of biopsies (red arrows) revealed the presence of different ALK secondary mutations throughout the treatments. Tumor growth observed during ceritinib treatment is consistent with disease progression. WT, wild-type.

note, 2 of the patients had crizotinib-resistant mutations reported differential activity of some of these ALK inhibitors before enrolling on ceritinib (MGH011; Fig. 6B; and JFCR021) depending on the resistance mutations present within the that our laboratory studies suggested would be sensitive to ALK TK domain (6 , 11 ). In an ongoing early-phase clinical ceritinib. In the ceritinib-resistant cancers, those mutations study, ceritinib has exhibited dramatic activity in patients were no longer detected, but the G1202R mutation emerged with ALK-rearranged NSCLC ( 13 ). ( Fig. 6B ). These fi ndings are consistent with preclinical studies In these studies, we fi nd that ceritinib is a more potent presented in this article demonstrating the activity of ceritinib ALK inhibitor than crizotinib, and has marked activity in against G1269A and S1206Y crizotinib-resistant mutations, crizotinib-naïve models of ALK -positive NSCLC, including and its lack of potency against the G1202R mutation. H2228, H3122, and Ba/F3 cell lines in vitro and MGH006 primary explants in vivo . To better characterize the activity of ceritinib in crizotinib resistance, we developed a variety of DISCUSSION crizotinib-resistant models, including cell lines derived from Since its approval in the United States in 2011, the ALK biopsies from patients whose cancers had developed resist- inhibitor crizotinib has emerged as a standard of care for ance to crizotinib in the clinic. These models harbored differ- patients with advanced NSCLC harboring the ALK fusion ent resistance mechanisms, including various ALK resistance oncogene. Unfortunately, as has been observed with other mutations. The activity of ceritinib varied depending on the targeted therapies, the emergence of resistance has ultimately specifi c ALK resistance mutation. For example, in Ba/F3 limited the benefi t of this therapy. Next-generation ALK models, ceritinib was highly active against L1196M, G1269A, inhibitors (ceritinib, CH5424802, ASP3026, AP26113, and S1206Y, and I1171T EML4–ALK mutants, and less active X-396) have been developed with the hope that they may against the less common mutations C1156Y, G1202R, 1151T- overcome acquired resistance to crizotinib. We previously ins, L1152P, and F1174C. It is notable that in the phase I

670 | CANCER DISCOVERYJUNE 2014 www.aacrjournals.org

Ceritinib and Crizotinib Resistance RESEARCH ARTICLE

study of ceritinib, fi ve of 19 crizotinib-resistant tumors har- procedures were conducted under an Institutional Review Board bored resistance mutations at residues 1196, 1269, and 1206, (IRB)–approved protocol. Cells in pleural effusion were collected by with one tumor harboring both G1269A and 1151T-ins. The centrifugation at 440 × g for 10 minutes. After red blood cells were patients harboring these resistance mutations all exhibited lysed with the Red Blood Cell Lysis Solution (BioLegend), cells were signifi cant tumor shrinkage (13 ). grown in ACL-4 (Invitrogen) supplemented with 1% FBS or RPMI- 1640 supplemented with 10% FBS and 1× Antibiotic-Antimycotic. Importantly, as has been observed in the clinic, ceritinib After the cells started growing stably, clonal cell lines were also showed potent effi cacy in vitro and in vivo against a crizo- established. tinib-resistant tumor that did not harbor an ALK resistance H3122, H2228, A549, H460, H1299, HCC827, and H522 cell lines mutation or gene amplifi cation (Fig. 3B). Interestingly, the were provided by the Center for Molecular Therapeutics (CMT) patient-derived cell line also retained sensitivity to crizotinib at Massachusetts General Hospital (Boston, MA), which performs in vitro, demonstrating that these cells are still sensitive to routine cell line authentication testing by single-nucleotide poly- ALK inhibition. One potential explanation for this fi nd- morphism and short-tandem repeat analysis. BT-474, SKBR3, and ing is that, in the clinic, crizotinib fails to achieve tumor the ALK -positive patient-derived cell lines used in this study are from levels that completely inhibit ALK, and that tumor cells can the Engelman laboratory (Boston, MA) and have been previously survive through modest input from activation of bypass tested for mutation status to confi rm their authenticity. A549, H460, H1299, HCC827, H522, SKBR3, H2228, H3122, H3122 CR1, and tracks such as EGFR. However, these cells remain sensitive MGH021-4 cell lines were cultured in RPMI-1640 supplemented with to complete ALK inhibition. In the setting of a more potent 10% FBS. For survival assays, H2228 were cultured in 1% FBS. The ALK inhibitor, ALK is inhibited fully, abrogating the func- MGH045 cell line was cultured in ACL-4 supplemented with 1% FBS, tional role of bypass tracks and leading to the elimination and MGH051 and BT-474 were cultured in DMEM supplemented of tumor cells. It is also possible that this patient relapsed with 10% FBS. on crizotinib because of poor adherence to therapy or due to Mouse myeloma Ba/F3 cells were cultured in DMEM supple- a stromal contribution. Similar fi ndings were also observed mented with 10% FBS with (parental) or without (EML4–ALK) IL3 in the H2228 xenograft model that developed resistance to (0.5 ng/mL). cDNAs encoding EML4–ALK variant1 or EML4–ALK crizotinib in vivo , did not develop an ALK mutation, and was variant3 containing different point mutations were cloned into sensitive to ceritinib ( Fig. 5A ). These fi ndings may explain, retroviral expression vectors, and virus was produced as previously described ( 11 ). After retroviral infection, Ba/F3 cells were selected at least in part, the fi nding that ceritinib is highly active in in puromycin (0.5 μg/mL) for 2 weeks. IL3 was withdrawn from the crizotinib-resistant cancers with or without ALK resistance culture medium for more than 2 weeks before experiments. mutations. Crizotinib was purchased from ChemieTek, and ceritinib was pro- The initial interrogation of ceritinib-resistant patient vided by Novartis. Both were dissolved in DMSO for in vitro experi- biopsies supports the notion that ceritinib is able to effec- ments. Ceritinib was formulated in 0.5% methyl cellulose/0.5% Tween tively suppress many crizotinib-resistant mutations, but the 80 and crizotinib in 0.1 N HCl or 0.5% methyl cellulose/0.5% Tween G1202R and F1174V/C mutants are resistant to ceritinib. It 80 for in vivo studies. is noteworthy that in two cases, the crizotinib-resistant mutations, S1206Y and G1269A, were no longer observed in the Western Blot Analysis ceritinib-resistant biopsies in which the G1202R mutations A total of 5 × 105 cells were treated in 6-well plates for 6 hours with were observed (Fig. 6A). This suggests that predominant the indicated drugs. Cell protein lysates were prepared as previously clones with the S1206Y and G1269A mutations were sup- described ( 6 , 11 ). Phospho-ERK (T202/Y204), ERK, S6, phospho-S6, pressed by ceritinib, whereas much more rare clones with phospho-AKT (S473 and T308), AKT, phospho-ALK (Y1282/1283), and ALK antibodies were obtained from Cell Signaling Technology. G1202R mutations were selected by ceritinib. These fi ndings GAPDH was purchased from Millipore. give further support to the notion that there are multiple populations of resistant clones whose emergence is depend- Survival Assays ent on the selective pressure applied. Cells (2,000 or 5,000) were plated in triplicate into 96-well plates. Altogether, our in vitro and in vivo data, including cell line Seventy-two hours (48 hours for Ba/F3 cells and 7 days for MGH051) models established from crizotinib-resistant patient samples, after drug treatments, cells were incubated with a CellTiter-Glo demonstrate that the next-generation ALK inhibitor ceritinib assay reagent (Promega) for 15 minutes, and luminescence was is active against most crizotinib-resistant tumors. This is measured with a Centro LB 960 Microplate Luminometer (Berthold consistent with the marked clinical activity of ceritinib in Technologies). patients with ALK -positive NSCLC who progressed on crizotinib. As resistance to ceritinib has already been observed in In Vivo Effi cacy Study of Ceritinib the clinic, future studies will need to identify mechanisms of SCID beige mice for crizotinib-resistant H2228 xenograft tumor resistance to ceritinib other than mutations in the G1202 and models, nude mice for MGH006 primary explants and MGH045 F1174 residues to maximize the clinical benefi t afforded by cells were randomized into groups of 5, 6, or 8 mice with an average 3 next-generation ALK-targeted therapies. tumor volume of approximately 150 mm and received crizotinib or ceritinib daily treatments by oral gavage as indicated in each study. Tumor volumes were determined by using caliper measurements and × × METHODS calculated with the formula (length width height)/2. Cell Lines and Reagents In Vitro Enzymatic Assay All human lung cancer samples were obtained from patients with An enzymatic assay for the recombinant ALK kinase domain informed consent at the Massachusetts General Hospital (MGH) (1066–1459) was conducted using the Caliper mobility shift method- and the Japanese foundation for Cancer Research (JFCR), and all ology, using fl uorescently labeled peptides as kinase substrates. The

JUNE 2014CANCER DISCOVERY | 671

RESEARCH ARTICLE Friboulet et al.

Caliper assay was performed at 30°C for 60 minutes in a total volume Spirito Memorial Foundation for support of lung cancer research of 9 μL. The reaction was terminated by the addition of 16 μL of stop at MGH. solution [100 mmol/L HEPES, 5% (v/v) DMSO, 0.1% (v/v) coating reagent, 10 mmol/L EDTA, 0.015% (v/v) Brij 35]. After termination Grant Support of the reactions, the plates were transferred into the Caliper LabChip This work was supported by a grant from the NIH (5R01CA164273- 3000 workstation for analysis. 02 to A.T. Shaw and J.A. Engelman), by a V Foundation Translational Research Grant (to A.T. Shaw and J.A. Engelman) and by the NIH/ Analysis of ALK/ Ceritinib and ALK/ Crizotinib Costructures National Cancer Institute (R01CA137008 to J.A. Engelman). The The ALK/ceritinib costructure was determined by the soaking of 2 study was also supported by a grant from JSPS KAKENHI (25710015 mmol/L ceritinib into apo crystals grown in 0.2 mol/L sodium ace- to R. Katayama). tate trihydrate/20% PEG3350 using protein expressed and puriﬁ ed as previously described ( 18 ). The ALK/ceritinib ﬁ nal model determined Received November 5, 2013; revised March 12, 2014; accepted to 2.0 Å (PDB 4MKC on hold) was superimposed with the coordi- March 19, 2014; published OnlineFirst March 27, 2014. nates of the ALK/crizotinib costructure (PDB 2XP2) for analyses.

Patient Sample Analyses REFERENCES The patients with ALK-positive NSCLC with acquired ceritinib 1. Kwak EL , Bang YJ , Camidge DR , Shaw AT , Solomon B , Maki R , resistance underwent biopsy of their resistant tumors between et al. Anaplastic lymphoma kinase inhibition in non-small-cell lung January 2011 and September 2013. Standard histopathology was cancer. N Engl J Med 2010 ; 363 : 1693 – 703 . performed to confi rm the diagnosis of malignancy as previously 2. Soda M , Choi YL , Enomoto M , Takada S , Yamashita Y , Ishikawa S , described (6 ). The electronic medical record was reviewed retrospec- et al. Identifi cation of the transforming EML4-ALK fusion gene in tively to obtain clinical information under an IRB-approved protocol. non-small-cell lung cancer. Nature 2007 ; 448 : 561 – 6 . This study was approved by the IRB of MGH or the Cancer Institute 3. Koivunen JP , Mermel C , Zejnullahu K , Murphy C , Lifshits E , Holmes Hospital of JFCR. AJ , et al. EML4-ALK fusion gene and effi cacy of an ALK kinase inhibitor in lung cancer. Clin Cancer Res 2008 ; 14 : 4275 – 83 . Disclosure of Potential Confl icts of Interest 4. Doebele RC , Pilling AB , Aisner DL , Kutateladze TG , Le AT , Weic k- hardt AJ , et al. Mechanisms of resistance to crizotinib in patients M. Nishio has received honoraria from the speakers’ bureaus of with ALK gene rearranged non-small cell lung cancer. Clin Cancer Res Pfi zer and Chugia Pharmaceutical Co., Ltd. A.T. Shaw is a consult- 2012 ; 18 : 1472 – 82 . ant/advisory board member of Novartis, Pfi zer, and ARIAD. J.A. 5. Gainor JF , Varghese AM , Ou SH , Kabraji S , Awad MM , Katayama R , Engelman has received commercial research grants from Novartis et al. ALK rearrangements are mutually exclusive with mutations in and Sanofi -Aventis, and is a consultant/advisory board member EGFR or KRAS: an analysis of 1,683 patients with non-small cell lung of Novartis, Sanofi -Aventis, Chugia Pharmaceutical Co., Ltd., and cancer. Clin Cancer Res 2013 ; 19 : 4273 – 81 . Ventana Medical Systems, Inc. No potential confl icts of interest were 6. Katayama R , Shaw AT , Khan TM , Mino-Kenudson M , Solomon BJ , disclosed by the other authors. Halmos B , et al. Mechanisms of acquired crizotinib resistance in ALK- rearranged lung Cancers. Sci Transl Med 2012 ; 4 : 120ra117 . Authors’ Contributions 7. Choi YL , Soda M , Yamashita Y , Ueno T , Takashima J , Nakajima T , Conception and design: L. Friboulet, N. Li, P.-Y. Michellys, M.M. Awad, et al. EML4-ALK mutations in lung cancer that confer resistance to A.C. Pferdekamper, S. Kasibhatla, F. Sun, J.L. Harris, A.T. Shaw, ALK inhibitors. N Engl J Med 2010 ; 363 : 1734 – 9 . J.A. Engelman 8. Lovly CM , Pao W . Escaping ALK inhibition: mechanisms of and strategies to overcome resistance. Sci Transl Med 2012 ; 4 : 120ps122 . Development of methodology: L. Friboulet, R. Katayama, M.M. Awad, 9. Sasaki T , Koivunen J , Ogino A , Yanagita M , Nikiforow S , Zheng S. Kim, A.C. Pferdekamper, J. Li, S. Kasibhatla, F. Sun, S. Mahmood, W , et al. A novel ALK secondary mutation and EGFR signaling E.L. Lockerman, N. Fujita cause resistance to ALK kinase inhibitors . Cancer Res 2011 ; 71 : Acquisition of data (provided animals, acquired and managed 6051 – 60 . patients, provided facilities, etc.): L. Friboulet, R. Katayama, 10. Sasaki T , Okuda K , Zheng W , Butrynski J , Capelletti M , Wang L , et al. C.C. Lee, J.F. Gainor, A.S. Crystal, N.Yanagitani, A.C. Pferdekamper, The neuroblastoma-associated F1174L ALK mutation causes resist- F. Sun, X. Sun, S. Hua, P. McNamara, M. Nishio, A.T. Shaw ance to an ALK kinase inhibitor in ALK-translocated cancers . Cancer Analysis and interpretation of data (e.g., statistical analysis, biosta- Res 2010 ; 70 : 10038 – 43 . tistics, computational analysis): L. Friboulet, N. Li, R. Katayama, 11. Katayama R , Khan TM , Benes C , Lifshits E , Ebi H , Rivera VM , et al. C.C. Lee, J.F. Gainor, M.M. Awad, N.Yanagitani, A.C. Pferdekamper, Therapeutic strategies to overcome crizotinib resistance in non-small S. Kasibhatla, F. Sun, P. McNamara, J.L. Harris, A.T. Shaw, J.A. Engelman cell lung cancers harboring the fusion oncogene EML4-ALK. Proc Writing, review, and/or revision of the manuscript: L. Friboulet, Natl Acad Sci U S A 2011 ; 108 : 7535 – 40 . N. Li, R. Katayama, C.C. Lee, J.F. Gainor, A.S. Crystal, M.M. Awad, 12. Marsilje TH , Pei W , Chen B , Lu W , Uno T , Jin Y , et al. Synthesis, struc- J.L. Harris, A.T. Shaw, J.A. Engelman ture-activity relationships, and in vivo effi cacy of the novel potent and Administrative, technical, or material support (i.e., report- selective anaplastic lymphoma kinase (ALK) inhibitor 5-Chloro-N2-(2- ing or organizing data, constructing databases): R. Katayama, isopropoxy-5-methyl-4-(piperidin-4-yl)phenyl)-N4-(2-(isopropylsulf A.C. Pferdekamper, S. Mahmood, E.L. Lockerman, N. Fujita, onyl)phenyl)pyrimidine-2,4-diamine (LDK378) currently in phase 1 M. Nishio, J.A. Engelman and phase 2 clinical trials. J Med Chem 2013 ; 56 : 5675 – 90 . Study supervision: N. Li, S. Kasibhatla, M. Nishio, A.T. Shaw, 13. Shaw AT, Kim DW, Mehra R, Tan DS, Felip E, Chow LQ, et al. Ceritinib in ALK-rearranged non-small-cell lung cancer. N Engl J Med J.A. Engelman 2014;370:1189–97. Acknowledgments 14. Early Results Promising for LKD378 in ALK-positive NSCLC . Cancer Discov 2013 ; 3 : OF5 . The authors thank Thomas Marsilje, Celin Tompkins, and 15. FDA, Center for Drug Evaluation and Research . 2011 Application Auzon Steffy for expert technical assistance and input on the stud- Number: 202570Orig1s000; Reference ID: 3006911 . ies described in the article; Atsushi Horiike for helping to obtain 16. Cui JJ , Tran-Dube M , Shen H , Nambu M , Kung PP , Pairish M , et al. repeat biopsy samples; and Be a Piece of the Solution and the Evan Structure based drug design of crizotinib (PF-02341066), a potent

672 | CANCER DISCOVERYJUNE 2014 www.aacrjournals.org

Ceritinib and Crizotinib Resistance RESEARCH ARTICLE

and selective dual inhibitor of mesenchymal-epithelial transition 18. Lee CC , Jia Y , Li N , Sun X , Ng K , Ambing E , et al. Crystal structure of factor (c-MET) kinase and anaplastic lymphoma kinase (ALK). J Med the ALK (anaplastic lymphoma kinase) catalytic domain. Biochem J Chem 2010 ; 54 : 6342 – 63 . 2010 ; 430 : 425 – 37 . 17. Chand D , Yamazaki Y , Ruuth K , Schonherr C , Martinsson T , Kogner 19. Zhang S , Wang F , Keats J , Zhu X , Ning Y , Wardwell SD , et al. P , et al. Cell culture and Drosophila model systems deﬁ ne three Crizotinib-resistant mutants of EML4-ALK identiﬁ ed through an classes of anaplastic lymphoma kinase mutations in neuroblastoma. accelerated mutagenesis screen. Chem Biol Drug Des 2011 ; 78 : 999 – Dis Models & Mech 2013 ; 6 : 373 – 82 . 1005 .

JUNE 2014CANCER DISCOVERY | 673

The ALK Inhibitor Ceritinib Overcomes Crizotinib Resistance in Non−Small Cell Lung Cancer

Luc Friboulet, Nanxin Li, Ryohei Katayama, et al.

Cancer Discovery 2014;4:662-673. Published OnlineFirst March 27, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/2159-8290.CD-13-0846

Supplementary Access the most recent supplemental material at: Material http://cancerdiscovery.aacrjournals.org/content/suppl/2014/03/27/2159-8290.CD-13-0846.DC1 http://cancerdiscovery.aacrjournals.org/content/suppl/2021/03/04/2159-8290.CD-13-0846.DC2

Cited articles This article cites 18 articles, 8 of which you can access for free at: http://cancerdiscovery.aacrjournals.org/content/4/6/662.full#ref-list-1

Citing articles This article has been cited by 73 HighWire-hosted articles. Access the articles at: http://cancerdiscovery.aacrjournals.org/content/4/6/662.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerdiscovery.aacrjournals.org/content/4/6/662. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.